The disclosures of the following priority application(s) are herein incorporated by reference:
Japanese Patent Application Laid Open No. 09-210315 filed Aug. 5, 1997.
1. Field of Invention
The present invention relates to an alignment apparatus and method which permit mask patterns to be precisely aligned with pattern transfer objective regions on a substrate, in accordance with coordinates predicted on the basis of, for example, a statistical technique. More specifically, the present invention is concerned with an alignment apparatus and method suitable for use in an Enhanced Global Alignment (EGA) process.
2. Description of Related Art
Semiconductor devices and liquid crystal display devices are produced by apparatus and methods that employ a lithographic process in which a reticle pattern, i.e. a mask pattern, is transferred to successive shot regions on a wafer or a glass plate coated with a photo-resist. The lithographic process is performed using a projection optical system having a projection exposure device such as a stepper. This process requires a high degree of precision of alignment between the reticle patterns, through which the exposure is to be performed, and circuit patterns which have already been formed on the respective shot regions on the wafer.
For example, U.S. Pat. No. 4,780,617 (Japanese Patent Laid-Open No. 61-44429) discloses a method of Enhanced Global Alignment (EGA), which realizes a high degree of precision of alignment between the reticle patterns and circuit patterns. According to this method, coordinate positions of alignment marks (wafer marks), provided on preselected sample shot regions on a wafer, are measured and the results of the measurement are statistically processed to determine the coordinate positions of the shot regions.
The alignment in accordance with the EGA technique is a kind of fine alignment. In order that the fine alignment be executed satisfactorily, it is necessary that a search alignment be performed such that the wafer marks on the sample shots regions fall within the sensing coverage of the alignment sensor without fail.
FIG. 9 shows a wafer 51 on which alignment is performed in accordance with the conventional EGA technique. A multiplicity of shot regions are allocated to the surface of the wafer 51 at predetermined pitches in two orthogonal directions, which will be referred to as the X and Y directions, respectively. Preselected shot regions, e.g., three shot regions, among these shot regions are provided with different search alignment marks WX, WY and Wxcex8, respectively, roughly indicating the positions in the X, Y and rotational (xcex8) directions. Each of the other shot regions is provided with a wafer mark 55 indicative of a two-dimensional position and composed of X-axis wafer mark 54X and Y-axis wafer mark 54Y.
In the actual EGA process, the search alignment marks and the wafer marks are formed on the boundaries between adjacent shot regions. Such boundary regions are also referred to as xe2x80x9cstreet regions.xe2x80x9d In the EGA process, ten shot regions C1 to C10, which are distributed substantially at a constant pitch in the circumferential direction, are selected as sample shots regions from among all the shot regions.
As the first step of the process, measurements are sequentially executed by a search alignment sensor. More specifically, the Y coordinate of the Y-axis search alignment mark WY, the Y coordinate of the xcex8-axis search alignment mark wxcex8 and the Y coordinate of the X-axis search alignment mark WX are sequentially measured. A conversion parameter composed of a rotational angle and an offset, necessary for converting the sample coordinate system on the wafer 51 into a stationary coordinate system of a wafer stage, is determined based on the results of the measurement. Coordinate positions of the sample shots are then determined on the stationary coordinate system to thereby, complete the search alignment.
Thereafter, the wafer marks 55 on the sample shots C1 to C10 are successively moved into the sensing area of a fine alignment sensor so that the coordinates of these wafer marks 55 on the stationary coordinate system are measured. The coordinates of the shot regions on the stationary coordinate system are determined through a statistical processing of the results of measurement. Exposure is then performed for each of the shot regions while aligning the shot regions in accordance with the coordinate positions on the stationary coordinate system. It is thus possible to achieve a high degree of overlay accuracy through the EGA process.
The conventional EGA alignment essentially requires a search alignment in order to ensure that the wafer marks of the sample shots fall within the sensing area of the fine alignment sensor. For example, in the case of the wafer shown in FIG. 9, it is necessary that the wafer be moved to three different positions in order to measure the positions of the search alignment marks WX, WY and wxcex8. Subsequently, the wafer is moved to ten positions for the purpose of measuring the positions of the wafer marks 55 on the ten sample shots. Consequently, a considerably long time is required for the alignment. As a result, the throughput of the exposure is undesirably reduced.
Another disadvantage encountered with the known alignment process is that the area on each shot region available for the circuit pattern is limited. This is due to the necessity of using both search alignment marks and fine alignment wafer marks which are to be formed on the shot regions or on the street line regions which are defined between adjacent shot regions.
Further, a complicated positioning control is required for the wafer stage in the conventional EGA adjustment. The degree of freedom of correction is also limited because the position of each shot region is finally corrected by controlling the position of the wafer stage in accordance with the results of the fine alignment executed by using the EGA process.
The present invention provides an alignment apparatus and method for aligning the position of each of the shot regions on a wafer based on the results of a statistical processing of the measured positions of a preselected number of the shot regions on the wafer. The alignment of the present invention employs a reduced number of measurement marks and is performed in a shorter processing time, as compared with known techniques, without impairing the measuring accuracy.
According to one aspect of the present invention, there is provided an alignment apparatus and method for successively bringing a plurality of shot regions to be processed on a substrate into alignment with a predetermined reference position. The shot regions to be processed are arranged on the substrate in accordance with design array coordinates so that a mask pattern is transferred to each of the shot regions successively. In this apparatus and method, m measurement objective regions are selected from among the plurality of shot regions. The selected measurement objective regions are successively brought into a predetermined measuring area to measure the coordinates of the measurement objective regions. A statistical computation is performed on the measured coordinates of the selected measurement objective regions thereby computing a linear error of the actual coordinates of each of the shot regions on the substrate from the design array coordinates. Then, relative correction on the position to which the substrate is to be moved is performed in accordance with the computed linear error.
The alignment method of the present invention comprises: (1) selecting k pieces of a preparatory measurement objective region from among the m pieces of measurement objective region; (2) determining the array coordinates of the plurality of shot regions based on the outline of the substrate; (3) successively bringing the k pieces of preparatory measurement objective region into a predetermined measuring area in accordance with the determined array coordinates, thereby measuring the coordinate positions of the k pieces of preparatory measurement objective region; (4) processing the results of the measurement of the coordinate positions of the k pieces to calculate parts (e.g., parts of the offset and parts of the rotation) of the linear errors of the actual array coordinates of the plurality of shot regions with respect to the design array coordinates; (5) updating the measured coordinate positions in accordance with the results of the calculation; (6) successively bringing (mxe2x88x92k) pieces of the measurement objective region (B1 to B8) into the predetermined measuring area based on the array coordinates corrected in accordance with the computed linear errors, thereby measuring the coordinate positions of the (mxe2x88x92k) pieces of the measurement objective regions; and (7) performing a statistical computation on the updated coordinate positions and the measured coordinate positions of the (mxe2x88x92k) pieces of the measurement objective regions, thereby computing the linear error of the actual coordinates of each of the plurality of shot regions from the design coordinates.
The alignment method of the present invention has been accomplished with alignment sensors having wide sensing areas with which high degrees of accuracy can be achieved in the rough alignment (pre-alignment) process based on the object outline employed in wafer loader systems of exposure apparatuses. Thus, in the pre-alignment of the present method, the array coordinates of the shot regions to be processed are determined with such an accuracy that causes alignment marks attached to the measurement objective shot regions to fall within a sensing area sensible by an alignment sensor which is employed in the subsequent steps. It is therefore not necessary to employ a separate step of search alignment for the purpose of enabling the sensing area of the alignment sensor to encompass the alignment marks, as is required in the known prior art processes. Consequently, the number of the marks to be used can be reduced because the marks which are used for search alignment in the known processes can be eliminated.
However, if the alignment sensor is of an image processing type, there is a risk that errors are involved in the positions detected at a peripheral region of the sensing area due to distortion of the image forming optical system. Prolongation of the detection time, as well as an increase in the detection error, tends to occur even with other types of alignment sensors if the alignment marks are positioned in a peripheral region of the sensing area.
Therefore, in accordance with the method of the present invention, the positions of the k pieces of preparatory measurement objective regions are successively measured, and the parts of the linear errors, such as offset and rotation, are then determined. Thus, correction based on the determined linear errors has been effected on the determined coordinate positions when the positions of the remaining measurement objective regions are measured. Therefore, the alignment marks carried by these measurement objective regions are located in the vicinity of the center of the sensing area provided by the alignment sensor. Consequently, the detection error can be reduced and, since the alignment marks are detectable from a smaller area around the center of the sensing area, the time required for the detection is also reduced.
Further, the results of the measurement of the positions of the k pieces of preparatory measurement objective regions obtained through the correction performed in the fourth step, are used in the computation of the final linear error. This permits the number of the alignment marks of the measuring objects to be reduced almost to the same number as that employed in the fine alignment step of the known process. Consequently, the number of the marks to be measured and hence, the processing time inclusive of the measuring time, are reduced.
Preferably, the fourth step is executed such that the k pieces of preparatory measurement objective regions are again brought into the predetermined sensing area, based on the array coordinates corrected based on the calculated linear error, so that the coordinate positions of the preparatory measurement objective regions are measured again to update the data concerning the coordinate positions. This repeated measurement offers a higher accuracy of the measurement because the second measurement is performed while the alignment marks on the k pieces of preparatory measurement objective regions are located sufficiently close to the center of the sensing area presented by the alignment sensor.
In accordance with another aspect of the invention, the alignment method of the first aspect employs, as the k pieces of preparatory measurement objective regions, a first preparatory measurement objective region and a second preparatory measurement objective region. In addition, the third and fourth steps include the steps of: bringing the first preparatory measurement objective region into the predetermined measuring area based on the array coordinates determined in the first step, thereby measuring the coordinate position of the first preparatory measurement objective region; processing the results of the measurement of the coordinate position to compute an offset which is a component of the linear error of the actual array coordinates of the plurality of shot regions; correcting the array coordinates in accordance with the computed offset; measuring the coordinate position of the first preparatory measurement objective region again based on the array coordinates corrected in accordance with the computed offset; bringing the second preparatory measurement objective region into the predetermined measuring area based on the determined array coordinates determined, thereby measuring the coordinate position of the second preparatory measurement objective region; and processing the results of the measurement of the coordinate position of the second preparatory measurement objective region to compute an angle of rotation which is another component of the linear error of the actual array coordinates of the plurality of shot regions; correcting the array coordinates in accordance with the computed rotation angle; and measuring the coordinate position of the second preparatory measurement objective region again based on the array coordinates corrected in accordance with the computed rotation angle.
These features achieve the correction of offset and rotation, through the measurement of positions of two preparatory measurement objective regions. Such a correction improves the accuracy of measurement of the remaining measurement objective regions.
In the alignment method of the present invention, the linear error calculated in the fifth step may be corrected through a control of the position of the mask pattern. Exposure apparatus perform a demagnification projection at an image contracting ratio of approximately xc2xc or ⅕. Thus, if the laser interferometer associated with the mask and the laser interferometer associated with the substrate have almost the same detection accuracy, an error of detection performed by the laser interferometer associated with the mask appears on the substrate in a magnitude reduced at the same ratio as the demagnification performed by the projection exposure apparatus. Therefore, when the projection is conducted in a manner to demagnify the mask pattern image, it is preferred that the correction of position be effected by controlling the position of the mask pattern so that misalignment can be effectively reduced.
The correction of misalignment through the control of the mask pattern position offers a higher speed of alignment as compared with the case where the correction is effected merely by the control of position of the substrate. Furthermore, the accuracy of correction of non-linear errors can be improved corresponding to the projection demagnification, when the correction is performed through the control of the mask pattern position.
The above and other objects, features and advantages of the present invention are described in or are apparent from the following description of the preferred embodiments.